The evolution of milk yield during lactation : a predictive model in the dairy cow

Abstract

The objective of this study was to build an empirical model for the prediction of the evolution of milk yield during lactation for correctly fed cows without any significant disease. It was elaborated from a set of 877 lactations coming from 5 experimental herds of the INRA, representing a broad range of situations (yield level, calving period, age...) which provided distinct samples for the development and validation of the model. The final proposed model integrates the effects of lactation stage, gestation stage and season. Input variables are calving date, parity, insemination date and yield potential. The latter is estimated from a yield measured between weeks 2 and 8 of lactation. The final model provides a very good fit of individual production curves, even when they present an unusual shape. The quality of prediction on an individual scale is limited: 29% of lactations of the validation sample present an average variation with the model higher than 2 kg/d. By contrast, with groups of cows the prediction becomes excellent: with 20 cows the average variation with the model exceeds 1 kg/d in only 2% of the cases. This model is thus an effective tool for the prediction of a reference of production on the herd scale. Its parameters are specified in the appendix.